You are not logged in.
How to specify the installation of a specific version of CUDA and cudnn?
I need to run a project using CUDA 10.1 while my ARCH is 12.6, which installed by "sudo pacman -S cuda cudnn".
I search the NVIDIA official website and find out that there only has version for Ubuntu. Does it suitable for ARCH?
Offline
If you need a specific version it's probably easier to do via user-level tools (as opposed to a system package manager) like anaconda or others.
Offline
If you need a specific version it's probably easier to do via user-level tools (as opposed to a system package manager) like anaconda or others.
My Anaconda version and CUDA version is right, but when I run this project in my bash and it shows"ValueError: Unknown CUDA arch (8.9) or GPU not supported", which seems problem happened on the system level.
Thank for your reply! I will try to solve it.
Offline
How to specify the installation of a specific version of CUDA and cudnn?
I need to run a project using CUDA 10.1 while my ARCH is 12.6, which installed by "sudo pacman -S cuda cudnn".
I search the NVIDIA official website and find out that there only has version for Ubuntu. Does it suitable for ARCH?
The biggest headache for me is that I don't know how to install an old version of CUDA on arch linux, AUR doesn't have an old version of CUDA, and the NVIDIA official website only has an Ubuntu version.
Offline
Yeah what I meant with my previous reply is that you could probably create a anaconda environment where the development dependencies of this project are all pinned to the fitting versions.
Offline
For the Record : AUR has many old versions of cuda, but 10.1 is not present.
Disliking systemd intensely, but not satisfied with alternatives so focusing on taming systemd.
clean chroot building not flexible enough ?
Try clean chroot manager by graysky
Offline